AI delivered on its promise. That's the problem.
If one unit of development now gets done five times faster, how do we deal with the five units of alignment and management we have to do?
Coding isn't the bottleneck anymore - you are.
My calendar used to have gaps. Not a lot (I was busy), but there were these pockets where I'd put on headphones, open vscode, and disappear into a problem for two or three hours. That was the good stuff. That was the reason I went into software development and data engineering.
My calendar doesn't have gaps anymore.
Where did the time go?
Not because I'm working more hours. I'm not. I work about the same as I always did. But something shifted in what fills those hours, and it happened so gradually that I didn't notice until I was already inside it.
Here's what happened: AI got pretty damn good at the things I used to spend my day doing. Using Claude, Gemini or (insert your favourite LLM of the day) I can write code faster than I ever have. I can analyse data in a fraction of the time. I can produce more, ship more, move through technical work at a pace that would've seemed nuts two years ago. On pure coding tasks, our team measured something like a 5x speed improvement.
But here's the number nobody talks about: total project time only improved 1.5x.
Read that again. Five times faster at the work. 1.5 times faster at the project. Where did the other 3.5x go?
I'll tell you where it went. It went into our calendars.
More work gets done = more coordination required
Every unit of code that gets written faster creates a new unit of coordination work. A feature that used to take a week now takes a day — which means the questions that used to come once a week now come every day. Are we aligned on the approach? Does this match what the other team expects? Who's reviewing this? Does the PM know the timeline moved up? Can we sync on priorities now that everything's moving faster?
The development sped up. The alignment didn't. The planning didn't. The context-sharing and the follow-ups and the "just want to make sure we're on the same page" conversations — none of that got faster. If anything, it got worse, because now there's more output to coordinate around.
AI made each of us faster. Nobody made the space between us faster.
Here's the part that nobody's writing about yet.
I didn't just get more productive. I got quietly, gradually, irreversibly promoted — into a job I never applied for.
I used to be a developer who sometimes coordinated. Now I'm a coordinator who sometimes develops. The deep work, the flow states, that feeling of riffing on a half-formed idea with a coworker — that got compressed into a sliver of my day. What expanded to fill the space is logistics. Planning. Status updates. Alignment. More alignment. The word "alignment" now takes up more of my week than code does.
And the cruel part is that AI did this to me by making me better at the thing I actually enjoyed.
The better I got at coding, the less coding became my job. The more I could produce, the more I needed to coordinate that production. Every hour AI freed up got immediately consumed by the organizational overhead of operating at this new speed.
A prison of my own making. Built from my own productivity gains.
People keep measuring AI productivity the wrong way. They measure the task: "this coding task took 4 hours, now it takes 45 minutes." And that's real. I'm not disputing that.
But nobody works in tasks. People work in weeks. You have a 40-hour work week, and coding was only ever a fraction of it. Speed up that fraction by 5x and the total improvement is bounded by everything else — the meetings, the reviews, the decisions, the alignment, the re-alignment after the alignment didn't stick.
When we measured one of our projects end to end, coding was 5x faster. The project was 2x faster. The gap between those numbers is where the actual story is, and almost nobody is looking at it.
The hollowness of status updates
My calendar is full. I am not alone. I am in rooms with people all day.
But something is missing, and it took me a while to name it.
It's not loneliness. It's hollowness. I'm surrounded by people but we're doing logistics, not thinking. We're distributing information, not solving problems together. The meetings multiplied but the collaboration — the real kind, where two people are half-confused and building an idea together in real time — that got squeezed out.
The calendar is full. The work is hollow.
The Neverending Story (novel), Bastian can make an arbitrary number of wishes on AURYN, but he loses a memory every time he does so.
Optimising for speed
Everyone building AI tools right now is optimizing execution speed. That makes sense — engineers see an engineering bottleneck and they build for it. Faster code generation. Better autocomplete. More automation.
But the bottleneck moved. The constraint isn't "how fast can one person produce work" anymore. The constraint is how teams align, share context, build trust, follow through, and make decisions together. And that work is — I think — stubbornly, structurally human.
Not because nobody's tried to automate it. But because coordination isn't a convergent problem. Code has a correct output for a given input. Alignment doesn't. Alignment is "do we all actually mean the same thing when we say ready?" It's navigating ambiguity between people with different contexts, different incentives, different mental models. AI is extraordinary at convergent problems. It may be fundamentally limited at divergent ones.
I don't have proof of that. It's a hypothesis. But I've watched AI transform every part of my work except the part that now is my work, and that pattern feels like a signal, not a coincidence.
I don't have an answer
I don't have a neat conclusion here. I don't have a product to pitch or a framework to offer or a "5 ways to fix this" list.
I just have an observation that I can't shake: AI made us individually faster, and the thing that actually slowed down — the human, messy, unglamorous work of staying aligned — just sits there, growing, unaddressed, taking up more of the calendar every week.
We optimized the wrong bottleneck. And now we get to live in the calendar that created.